Journal
ENERGY ECONOMICS
Volume 68, Issue -, Pages 77-88Publisher
ELSEVIER
DOI: 10.1016/j.eneco.2017.09.010
Keywords
Oil prices; Forecasting; Least Absolute Shrinkage and Selection Operator (LASSO); Mean Squared Prediction Error (MSPE); Success ratio
Categories
Ask authors/readers for more resources
This paper identifies factors that are influential in forecasting crude oil prices. We consider six categories of factors (supply, demand, financial market, commodities market, speculative, and geopolitical) and test their significance in the context of estimating various forecasting models. We find that the Least Absolute Shrinkage and Selection Operator (LASSO) regression method provides significant improvements in the forecasting accuracy of prices compared to alternative benchmarks. Relative to the no-change and futures-based models, LASSO forecasts at the 8-step ahead horizon yield significant reductions in Mean Squared Prediction Error (MSPE), with MSPE ratios of 0.873 and 0.898, respectively. We also document substantial improvements in forecasting performance of the factor-based model that employs only a subset of variables chosen by LASSO. Finally, the time-varying nature of the relationship between factors and oil prices is used to explain recent movements in crude oil prices. (C) 2017 Elsevier B.V. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available